Monday, June 01, 2009

Mark Penn: sloppy with data?

Mark Penn biweekly "microtrends" column in today's Wall Street Journal surveys research showing "that women are succeeding in an ever-widening range of areas, while there is a statistically significant and growing group of guys who are just not going to make it." The findings are broadly familiar: men are likelier than women to be overweight, alcoholic, imprisoned, felled by heart attacks; they are underperforming women academically and increasingly in the workplace.

While there's nothing controversial in the column, a couple of errors and rhetorical sleights of hand made me wonder just what kind of value Hillary Clinton got from Penn as data guru.

Here's the one out-and-out error:
Almost three-quarters of men are obese, compared to fewer than two-thirds of women.
Three quarters of men are obese? Well, no. In the chart Penn links to (from the annual report of the Dept. of Health and Human Services) , the category from which his stat is drawn is labelled "Overweight (includes obesity)." "Overweight" is defined as a body mass index at or over 25 kilograms/meter; "obese" is at or over 30. We don't know what proportion of the total are obese -- or in fact whether there may be more obese women than men (and conversely, more merely overweight men than women).

The actual figures for the combined category (from 2003-2006) are 72.6% for men, 61.2% for women. I'm not sure what Penn gains by indeterminate rounding. At first I thought he was exaggerating to strengthen his case, but in fact his vague formulation understates the spread.

Next, this:
In December 2007, before the economic crisis set in, unemployment rates were roughly equal, but this time it's the guys who have been laid off in greater numbers. The unemployment rate surged from 4.4% to 7.2% for men, but from 4.3% to only 5.9% for women.
Those end-point unemployment figures sound a bit low...turns out they're from December 2008. It might make sense to mention that.

One more instance. There's nothing technically inaccurate here, but the data pulls against the implied conclusion in a way that should be acknowledged:
More women in the workplace, more female-led families and more sexual equality in general might be expected to cause women to pick up men's bad habits, or be exposed to greater dangers. But for the most part, that hasn't happened.

In fact, men continue to outstrip women in most of the downers of life. A record 1.5 million men are in the clink, up 600,000 in the last decade. Only 115,000 women have committed crimes worthy of jail -- and while that's doubled in the last 10 years, more men continue to wind up in jail by a ratio of nearly 15-to-1.
In other words, the number of men incarcerated increased 67% in a decade, while the number of women incarcerated increased 100%. Not exactly a roaring success story for American men. But it does look like women have "picked up men's bad habits" on this front. Their percentage of the overall prison population has risen.

I'll confess to feeling a bit petty in this post. But the column does make me wonder how rigorous Penn was in Microtrends -- or in his work for political and corporate clients. A look back at his WSJ opus might be in order.

No comments:

Post a Comment